analysis. The data essentially fit the normal distribution line, but show a trend to be above the

line for higher cumulative probabilities and below the line for lower cumulative probabilities. We

conclude that the data are approximately normally distributed.

Figure 5 suggests that the fixed monitor data are also normally distributed. Note that the data in Figure

5 are somewhat more randomly distributed on each side of the normal line, with fewer "runs" or

continual observations on one side or the other of the normal line. This graphically based determination is

subjective. To lessen subjectivity, tests as discussed above can be used (Kolmogorov-Smirnov, Lillefore's,

etc.), but the analyst is often forced to use whatever data are available.

Because the data were collected for another study and not specifically for monitor verification, the

transect locations did not coincide exactly with the monitor location. For our comparison, all transects

that were within 3.5 km of the fixed monitor station were selected. The number of samples varied with

transect mile and date. The May samples had five or seven evenly spaced measurements at a constant

depth of 4.6 m. July samples had multiple depths and five to seven sample locations. The calculations

of the differences, the square of the differences, and the totals of the two sites are depicted in Table 2.

18 May 95

119.9

1.7

2.9

25 May 95

121.2

2.6

6.8

25 May 95

121.6

1.7

2.9

25 May 95

122.1

0.9

0.8

25 May 95

119.9

3.6

13.0

27 Jul 95

121.2

2.3

5.3

27 Jul 95

121.6

4.1

16.8

27 Jul 95

122.1

3.4

11.6

7